Skip to content

Cloud VPS · Crypto trading bots · Gate

A VPS for Gate trading bots, in Tokyo.

Gate's API answered our Tokyo server in 4.7 ms, against 80 ms from Singapore (measured 30 October 2026). Linux VPS in Tokyo from $9.90 a month, with a dedicated IPv4 address you can lock your API key to.

Order in TokyoSee the measurements
  • 4.7 ms to Gate
  • from $9.90/mo
  • dedicated IPv4
  • Linux or Windows
  • full root

Latency measured 30 October 2026 from our own servers. Prices read from the catalogue the cart charges from.

Rendering of a network switch with three patch leads, one lit4.7 ms to GateTokyo

Step 01

Choose the city

Both cities we measured from, with the round trip to Gate's API and the lowest price in each. The faster one is open first.

Steps 02 to 05

Build your server in Tokyo.

The line and the size first, then the system, then the add-ons. Ubuntu 24.04 with Docker is selected for you; Windows appears on the sizes that carry it. The summary shows the price the cart charges.

Ordering in Tokyo, JapanChange city

Step 02

Pick a line

The most cores, RAM and disk per dollar in the catalogue, with a 32 TB monthly allowance.

Step 03

Choose a size

Sizes are fixed shapes, not sliders. Changing size later is a support request, and a smaller size usually means a new server because a disk cannot shrink.

Step 04

Choose an image

Clean vendor images, nothing preloaded that you did not ask for. You can rebuild onto any other image from the panel at any time.

Every image is the vendor's own, unmodified. If you want the machine configured before you log in, add a start-up script below.

Step 05

Choose a term and add-ons

Terms are prepaid: the longer the term, the lower the monthly rate.

Billing term

$18.90 charged every month. Switch to annual at any renewal.

How many

One machine. Each one you add is built separately, with its own address and root password.

1

Not available on BD-8 in Tokyo: nightly backups cannot be switched on for this machine. Take your own backups off the box, or pick a machine that has them.

The form shows the smaller sizes of each line. Every size in Tokyo

Our suggestion: 2 GB for a bot with its exchange library and a small database; 1 GB runs one light bot.

Measured

Gate from Tokyo and from Singapore.

Gate has not said where its matching engine runs; our measurement points to Tokyo.

Round-trip time in milliseconds from our servers to Gate's API
Exchange APIFrom TokyoFrom Singapore
GateV4 API4.7 msfastest80 ms
  • Measured 30 October 2026 from our own servers in Tokyo and Singapore: the best of nine HTTPS requests to the exchange's public ping or time call, on one open connection, so the figure is the round trip to the exchange's API without the connection set-up.
  • A first connection showed about 2 ms everywhere: that is the content network's edge in the same city, not the exchange, so we do not print it as the exchange's latency.
  • We measured from 3 data centres in Tokyo and 2 in Singapore on the same day and print the slowest of them. The other data centres we sell in the same city were not measured and can differ by a few milliseconds. Your own bot adds its processing time, and routes change; measure from the server you order before you rely on a figure.

Setup

From order to a running bot.

Linux over SSH, or Windows over Remote Desktop. Steps marked as our advice are ours; the rest is how the systems and exchanges work.

  1. Step 1

    Connect

    Your service page shows the address and the ssh command once the server is online.

    ssh root@YOUR_SERVER_IP
  2. Step 2 · our advice

    Keep the clock exact

    Signed API requests carry a timestamp the exchange checks, so a drifting clock gets requests refused. Check that time sync is on.

    timedatectl status | grep -i synchronized
  3. Step 3 · our advice

    Run the bot as a service

    A systemd service starts the bot at boot and restarts it if it stops; tmux is enough while you are testing.

    systemctl enable --now mybot.service journalctl -u mybot -f
  4. Step 4

    Lock the API key to this server

    Every server has its own IPv4 address that stays with it, so it can be the only address allowed to use your key.

    curl -4 -s https://ifconfig.me

Bots

The bots people run on it.

Install commands from each project's own documentation. The form opens on Ubuntu 24.04 with Docker installed, which both open-source bots need. Bots sold as a website, where you only paste API keys, run on that company's servers and need no VPS.

  • Freqtrade

    Open-source strategy bot in Python, with backtesting and a web interface.

    mkdir ft_userdata && cd ft_userdata/ curl https://raw.githubusercontent.com/freqtrade/freqtrade/stable/docker-compose.yml -o docker-compose.yml docker compose pull docker compose run --rm freqtrade create-userdir --userdir user_data docker compose run --rm freqtrade new-config --config user_data/config.json docker compose up -dFreqtrade: Docker quick start

  • Hummingbot

    Open-source market-making and arbitrage bot, run from a terminal client.

    git clone https://github.com/hummingbot/hummingbot.git cd hummingbot make setup make deployHummingbot on GitHub

  • CCXT

    The library most self-written bots use to talk to exchanges, in Python, JavaScript and more.

    pip install ccxtCCXT on GitHub

Other exchanges

Trading somewhere else too?

The same two cities, measured the same way; each page gives the date of its own measurement.

Every exchange in one table

Questions

What Gate traders ask before they move a bot to a VPS.

Which server location is best for Gate?
Tokyo. On 30 October 2026 Gate's API answered our Tokyo server in 4.7 ms and our Singapore server in 80 ms. Gate has not said where its matching engine runs; our measurement points to Tokyo.
Where are Gate's API servers located?
Gate has not published where its servers run. On 30 October 2026 its API answered our Tokyo server in 4.7 ms and our Singapore server in 80 ms, so Tokyo is the city to put a Gate bot in.
Will Gate's API work from your servers?
It did from every server we measured in Tokyo on 30 October 2026: each request was answered with HTTP 200. Each exchange decides who may trade from which country, so check Gate's terms for where you live.
How much server does a Gate bot need?
Most bots are light: 1 GB of memory runs one, 2 GB leaves room for its exchange library and a small database. The latency comes from the city, not the size, so a bigger plan does not make orders faster.
Can I run my bot on Windows?
Yes. Windows Server with its licence included, on sizes from 2 vCPU and 4 GB, from $30.90 a month in Tokyo and Singapore. Choose it in the system step. For the same bot, Linux sizes cost less.
Can a server in Tokyo or Singapore let me use an exchange where it is restricted?
No. Exchanges decide who may trade by the account holder, not by the server, and using a server to get around that breaks their terms and can close the account. Use a server for speed, from a country where the exchange serves you.
Will my bot keep running when I close SSH or Remote Desktop?
Yes, if it runs as a service: on Linux a systemd unit or Docker keeps it going after you log out and starts it again after a reboot. On Windows, close the Remote Desktop window rather than signing out; signing out ends the programs you started.
Is my API key safe on a VPS?
The server is yours alone, with root or Administrator access only for you. Give the key trading rights without withdrawal, lock it to the server's IPv4 address, and keep it in a file only your bot's user can read, not in the code.
Is PartVPS part of Gate?
No. We are an independent hosting company. We sell the server; your account, API keys and trading stay between you and Gate.

Binance, Bybit, OKX, Bitget and Hyperliquid belong to their owners. PartVPS is an independent hosting company, not affiliated with any exchange; we sell the server, not trading services or advice. Each exchange decides who may use it from where: check its terms for your own country.

Keep reading

Related

BD-8

$18.90 due today